Discussion with the Scientists: Unlocking Histopathology Real-World Data for Precision Oncology

We sat down with Head of BioPharma Product, Ryan Leung to discuss the potential for AI-powered pathology to boost Real-World Data to new heights.

In an era where precision medicine is critical to improving patient outcomes, RWD offers invaluable information on population diversity, treatment pathways, and long-term safety and effectiveness of drugs in the real-world setting. However, taking advantage of the vast amounts of data available requires innovative technologies, partnerships, and solutions to maximize actionable insights.

In this discussion, Ryan explains how new types of real-world data are enabling deeper, more personalized insights into patient health and treatment efficacy. Further, he discusses the unique challenges of histopathology data and the opportunity for digital pathology and AI platforms to drive the next frontier in RWD.

If you’re interested in learning more about Real-World Data, you can meet with Ryan and other PathAI team members at the upcoming ORIEN Scientific Summit on October 28-30 in Tampa Bay, FL. Learn more >>

Kaylee Mueller: Can you describe what Real-World data is and how it is collected and utilized?

Ryan Leung: Real-world data (RWD) are data relating to patient health status and/or the delivery of health care routinely collected from a variety of sources. Examples of RWD include data derived from electronic health records, medical claims data, data from product or disease registries, and data gathered from digital health technologies.

Traditional clinical trials, such as randomized controlled trials (RCTs), can take a long time and typically enroll a narrow patient population that represents a fraction of the patients who will likely receive the product in the real world. RWD can generate supplemental evidence that reflects a larger and more diverse patient population than is included in clinical trials and address timely clinical questions.

KM: How is RWD leveraged specifically for oncology research and drug development?

Ryan Leung: RWD has fueled tremendous advances in cancer care and research; transforming data captured during routine care into actionable evidence, improving patient outcomes, supporting FDA regulatory approvals, delivering clinical insights, and informing healthcare policy.

The pairing of clinical RWD; such as patient demographics, treatment regimens, and outcomes; with genomic RWD; which includes information on a patient's gene mutations, molecular signatures, and biomarkers; has been particularly transformative. Clinico-genomic RWD has helped answer questions such as:

• What is the real-world prevalence of actionable biomarkers across tumor types?
• How effective is a drug across different genomic subtypes of cancer? Different patient populations?
• Which study sites are best suited to efficiently recruit and retain clinical trial patients?
• How do geographic, socioeconomic, and demographic factors influence patient access to treatment?

As a result, pharmaceutical companies, diagnostic laboratories, healthcare providers, and regulatory agencies are increasingly relying on RWD to complement traditional data sources to make faster, data-driven decisions across the healthcare continuum.

KM: What is the role of histopathology in oncology RWD?

Ryan Leung: Despite the impact that clinico-genomic RWD has had in oncology, one critical source of information remains underutilized in the RWD landscape: histopathology (the branch of pathology involving the microscopic analysis of intact tissue specimens). Accurate and reproducible pathological analysis is essential to ensure that patients receive the right diagnosis and the right therapy. Additionally, histopathology plays a central role across the drug development life cycle, ranging from translational research (e.g., pharmacodynamic effects of drugs on cells), clinical development (e.g., assessment of histological endpoints) and commercialization (e.g., companion diagnostics).

Yet, histopathology’s contributions to RWD remain limited because most pathology has been solely dependent on human observations which are often subjective, not quantitative, and difficult to standardize and scale across large populations. The inability to provide quantitative, reproducible, structured, and scalable measurements creates a gap, limiting the unique multi-scale tissue and cellular-level data from pathology samples from contributing to multi-modal RWD analyses and precision medicine strategies.

KM: How can we overcome barriers to integrating histopathology with RWD?

Ryan Leung: These challenges make it clear that we need both new ways to visualize histopathology samples as well as new tools to analyze them to truly unlock the vast amount of information they contain. Enter digital pathology and artificial intelligence (AI). Advances in whole slide scanning and image processing have enabled a conversion from traditional glass slides to digitized slide images. Meanwhile, advances in machine learning have enabled AI systems to perform at, or better than, human level for a wide variety of fundamental tasks.

At PathAI, we bring together large, diverse datasets and cutting-edge computational tools to develop AI products that aim to extract quantitative, structured insights from pathology images that provide a level of scale and resolution that was not possible before. By automating the process of analyzing images, and standardizing the information extracted from each image, we enable histopathology data to be seamlessly integrated with other structured RWD.

KM: Can you elaborate how AI helps with histopathology RWD inclusion?

Ryan Leung: Through our PathExplore? Products, we transform unstructured pathology images into actionable pathology insights. Explore provides rich visualizations of the composition of each histopathology sample, including single-cell resolution classification of cells and tissues. Additionally, Explore extracts a structured panel of hundreds of quantitative metrics including both standard (e.g., density of tumor infiltrating lymphocytes) as well as novel (e.g., ratio of lymphocytes to fibroblasts near the epithelial-stromal interface) characteristics of each tissue sample.

Together these dense spatial maps and quantitative features can be used to refine existing diagnostic paradigms, identify completely new histologic biomarkers, and analyze alongside clinical and molecular data to build a more comprehensive, multi-modal understanding of disease.

KM: How can researchers access AI-powered pathology RWD? What are the different options for working with our partners??

Ryan Leung: PathAI partners with leading RWD providers to offer access to readily-available multimodal datasets that include clinical, molecular, and AI-powered pathology data from PathAI’s Explore products across most solid tumor types. We are also able to analyze data provided by researchers to enrich any other pathology datasets available. Lastly, PathAI provides access to Explore data linked to H&E images from The Cancer Genome Atlas (TCGA) for free to approved academic researchers. To learn more, reach out to us at [email protected] for more information!

PathExplore? is for research use only. Not for use in diagnostic procedures.?